Status of GitHub automatic tier1 testing
kevin.rushforth at oracle.com
Fri Oct 2 14:04:43 UTC 2020
Isn't this only run if the branch you push to has an open pull request?
The way I had imagined this working, is that it would run the test as
soon as you create a PR (even a Draft PR).
If it actually runs a build and tier 1 tests on every push to any branch
of everyone's personal fork, then it might be better to make it an
explicit "opt in".
On 10/2/2020 6:48 AM, Thomas Schatzl wrote:
> On 02.10.20 15:41, Robin Westberg wrote:
>> Hi all,
>> As you may have noticed when pushing changes to your personal forks
>> recently, GitHub will now automatically trigger basic tier1 testing
>> on the supported platforms (currently Linux, Windows and macOS, all
>> x64). If all goes well you may not even notice it as it defaults to
>> sending a notification only if anything should fail.
>> However, please note that the first version of this automatic test
>> execution contained a problem that could cause it to not detect
>> failing test cases. I know there have been a few test errors that
>> have gone by unnoticed due to this for which I’m sorry. The good news
>> is that the problem is now fixed starting with
>> - so make sure that your branch contains this change if you plan to
>> look at the test results!
>> Apart from this there are now no known issues with these pre-submit
>> tests, so if you run into any problem, please let me know. If these
>> tests turn out to be reasonably stable and trustworthy, the next step
>> will be to present a summary of the results in the body of PRs to
>> make them more visible.
>> Best regards,
> thanks for your effort.
> I would like to ask for a way to opt out - the public personal may
> receive unfinished work that may not even compile in a lot of cases.
> So this pre-checking seems like a waste of cpu time in a lot of cases
> and another notification I do not want in my inbox.
> After the fifth or so a day I'll probably just ignore it anyway.
> Also there are a significant amount of pushes that are done just
> fixing a typo in a comment or such, again wasting time.
> I can somewhat understand such testing for different situations.
More information about the jdk-dev